


Arizona - AI Water Chatbot
UX Designer | Conversational Design
3 Months
2024
My Role - UX Researcher
The Arizona Water Chatbot, developed by ASU's interdisciplinary team, helps with water and drought decisions amid a historic megadrought. It uses OpenAI’s technology for accurate, secure information and includes perspectives from Arizona’s 22 tribal communities.
The Arizona Water Chatbot, developed by ASU's interdisciplinary team, helps with water and drought decisions amid a historic megadrought. It uses OpenAI’s technology for accurate, secure information and includes perspectives from Arizona’s 22 tribal communities.
Why this chatbot and
What inspired this initiative?
Arizona faces severe water scarcity due to drought, climate change, and population growth. Inspired by the need for accessible and accurate information on water conservation, we created the Arizona Water Chatbot. This friendly technology equips residents and policymakers with the tools and knowledge to manage water resources sustainably. The chatbot aims to provide timely guidance, helping individuals and communities make informed decisions about effective water management and conservation efforts.
Arizona faces severe water scarcity due to drought, climate change, and population growth. Inspired by the need for accessible and accurate information on water conservation, we created the Arizona Water Chatbot. This friendly technology equips residents and policymakers with the tools and knowledge to manage water resources sustainably. The chatbot aims to provide timely guidance, helping individuals and communities make informed decisions about effective water management and conservation efforts.



Is the chatbot effective?
To address this question effectively, we conducted primary user research and user testing of the existing platform to understand Arizona residents' knowledge of water and technology and identify future opportunities for the product to meet their expectations. I led the research with 10 participants, all native Arizona residents, ensuring that product development aligned with their specific needs and perspectives. This approach allowed us to gather valuable insights and tailor the chatbot to effectively serve the community it was designed for.
To address this question effectively, we conducted primary user research and user testing of the existing platform to understand Arizona residents' knowledge of water and technology and identify future opportunities for the product to meet their expectations. I led the research with 10 participants, all native Arizona residents, ensuring that product development aligned with their specific needs and perspectives. This approach allowed us to gather valuable insights and tailor the chatbot to effectively serve the community it was designed for.






Research
User Demographics
Our primary user base selected for user testing consisted of individuals from Phoenix, Prescott, Northern Arizona, and Tucson.
We had students, long-term residents, business owners, and government officials participating in the study.
User Demographics
Our primary user base selected for user testing consisted of individuals from Phoenix, Prescott, Northern Arizona, and Tucson.
We had students, long-term residents, business owners, and government officials participating in the study.






Understanding the User
The primary research provided deep insights into users' mental models, revealing both strengths and areas for improvement in the product's design and functionality.
Below are what the user knows about certain topics related to Water in Arizona.
Understanding the User
The primary research provided deep insights into users' mental models, revealing both strengths and areas for improvement in the product's design and functionality.
Below are what the user knows about certain topics related to Water in Arizona.



Here are some interesting insights derived from the user research.



Testing the Product
The user testing sessions provided valuable insights into the issues users face when interacting with the chatbot and the types of questions they commonly ask. This feedback was critical to broaden the chatbot's scope, improving user experience, and ensuring the product meets users' needs effectively.
Below are the questions that users asked the chatbot.
Testing the Product
The user testing sessions provided valuable insights into the issues users face when interacting with the chatbot and the types of questions they commonly ask. This feedback was critical to broaden the chatbot's scope, improving user experience, and ensuring the product meets users' needs effectively.
Below are the questions that users asked the chatbot.



User - 1
"I liked the information the chatbot provided, but it would be more reassuring if the sources were mentioned. Knowing where the data comes from would make it feel more trustworthy. "
User - 1
"I liked the information the chatbot provided, but it would be more reassuring if the sources were mentioned. Knowing where the data comes from would make it feel more trustworthy. "
User - 2
"I often doubt the reliability of chatbot responses since I can’t verify their sources. It would be helpful to have a way to check the information’s origin to boost my confidence in its accuracy. "
User - 2
"I often doubt the reliability of chatbot responses since I can’t verify their sources. It would be helpful to have a way to check the information’s origin to boost my confidence in its accuracy. "
What did we do next?
A Chatbot Audit was conducted to identify design and functionality issues. Several factors made the chatbot less effective and trustworthy during its initial release. These were majorly surface, including issues.
A Chatbot Audit was conducted to identify design and functionality issues. Several factors made the chatbot less effective and trustworthy during its initial release. These were majorly surface, including issues.



If a User Error occurs, the chatbot does not allow users to easy escape or ability to edit their text.
If a User Error occurs, the chatbot does not allow users to easy escape or ability to edit their text.






- This section contains a significant amount of text. The details within this section may not be immediately visible to users.
- When a user clicks “New Chat” or closes the browser, the previous chat is cleared and lost.
- This section contains a significant amount of text. The details within this section may not be immediately visible to users.
- When a user clicks “New Chat” or closes the browser, the previous chat is cleared and lost.
These prompts would look unnoticed and not easily understandable by the user.
These prompts would look unnoticed and not easily understandable by the user.



Key Observations
A Chatbot Audit was conducted to identify design and functionality issues. Several factors made the chatbot less effective and trustworthy during its initial release. These were majorly surface, including issues.
A Chatbot Audit was conducted to identify design and functionality issues. Several factors made the chatbot less effective and trustworthy during its initial release. These were majorly surface, including issues.


Not following ASU Guidelines


More response time


Trust issue due to source credibility


Off-topic and straight responses
The challenges
Balancing the varying priorities of ASU faculty, Government agencies, and water management experts is challenging.
Faculties prioritize educational outcomes, government agencies ensure compliance with regulations, and water experts needed to emphasize the accuracy of information. Harmonizing these differing perspectives required careful coordination to ensure that the chatbot meets all stakeholder expectations while maintaining a cohesive design.
Faculties prioritize educational outcomes, government agencies ensure compliance with regulations, and water experts needed to emphasize the accuracy of information. Harmonizing these differing perspectives required careful coordination to ensure that the chatbot meets all stakeholder expectations while maintaining a cohesive design.
Adapting to constantly changing water management and data privacy regulations.
The chatbot responses had to comply with new laws and changes in existing regulations, which required regular updates and revisions to both the functionality and data handling processes.
The chatbot responses had to comply with new laws and changes in existing regulations, which required regular updates and revisions to both the functionality and data handling processes.
Maintaining ASU's brand guidelines while using a different mascot than the Sun Devil or Forks Up posed a challenge.
It was difficult to introduce a new character that still aligns with the university's visual and cultural standards. Ensuring the new mascot fits seamlessly within ASU's branding while being distinct required careful design choices and creativity to maintain brand consistency without diluting the university's branding.
It was difficult to introduce a new character that still aligns with the university's visual and cultural standards. Ensuring the new mascot fits seamlessly within ASU's branding while being distinct required careful design choices and creativity to maintain brand consistency without diluting the university's branding.
Iteration
Screenblueprint
Started with rough sketches and wireframing using the gathered data from research and user testing. We then addressed the identified problems, ensuring adherence to both ASU design guidelines and design principles.
Screenblueprint
Started with rough sketches and wireframing using the gathered data from research and user testing. We then addressed the identified problems, ensuring adherence to both ASU design guidelines and design principles.






High Fidelity Screens
The wireframes were developed into high-fidelity screens using Figma. Conversational design was also considered, as many responses were straightforward and difficult to read. This approach aimed to reduce cognitive load and enhance learnability for users.
High Fidelity Screens
The wireframes were developed into high-fidelity screens using Figma. Conversational design was also considered, as many responses were straightforward and difficult to read. This approach aimed to reduce cognitive load and enhance learnability for users.



Initial Design


The response format was structured in paragraphs, which posed a significant challenge for users trying to learn and understand the content efficiently. This format made it hard to quickly grasp key points and follow the information, leading to a frustrating and less effective learning experience.
The response format was structured in paragraphs, which posed a significant challenge for users trying to learn and understand the content efficiently. This format made it hard to quickly grasp key points and follow the information, leading to a frustrating and less effective learning experience.


Provided actionable buttons, tailored responses, and flexible formatting to improve learnability. This will simplify information delivery and reduce cognitive load, making it easier for users to understand and retain key points.
Provided actionable buttons, tailored responses, and flexible formatting to improve learnability. This will simplify information delivery and reduce cognitive load, making it easier for users to understand and retain key points.



Iterated Conversational Design


The paragraph format increased cognitive load by requiring users to skim through dense text to locate key information, making it more difficult to process and retain essential points effectively.

The paragraph format increased cognitive load by requiring users to skim through dense text to locate key information, making it more difficult to process and retain essential points effectively.
The paragraph format increased cognitive load by requiring users to skim through dense text to locate key information, making it more difficult to process and retain essential points effectively.

By incorporating actionable items that allow users to verify the source of information, we increase their trust in the product, ensuring transparency and credibility.
By incorporating actionable items that allow users to verify the source of information, we increase their trust in the product, ensuring transparency and credibility.
The redesigned screen will feature increased white space and include elements such as a help tip section in the bottom right corner to assist users in understanding the functions of actionable items. It will provide a concise overview of the product and include a dropdown section for information on how it works, frequently asked questions, resources, and history.
The redesigned screen will feature increased white space and include elements such as a help tip section in the bottom right corner to assist users in understanding the functions of actionable items. It will provide a concise overview of the product and include a dropdown section for information on how it works, frequently asked questions, resources, and history.






The sources section of the product has been redesigned, allowing users to easily view and click on them by pressing the actionable button for the response, enhancing the chatbot's trustworthiness.
The sources section of the product has been redesigned, allowing users to easily view and click on them by pressing the actionable button for the response, enhancing the chatbot's trustworthiness.
User Testing
Right after the designs were finalized and pushed to development, we conducted an eye-tracking test to understand how users interacted with the interface and to identify any visual elements that might attract or distract attention. This allowed us to gather precise data on gaze patterns, which informed our insights into the effectiveness of our design in guiding users through key tasks. By analyzing the eye-tracking data, we were able to refine our design further, ensuring that critical elements were positioned intuitively and that the overall user experience was optimized for ease of use and engagement.
Right after the designs were finalized and pushed to development, we conducted an eye-tracking test to understand how users interacted with the interface and to identify any visual elements that might attract or distract attention. This allowed us to gather precise data on gaze patterns, which informed our insights into the effectiveness of our design in guiding users through key tasks. By analyzing the eye-tracking data, we were able to refine our design further, ensuring that critical elements were positioned intuitively and that the overall user experience was optimized for ease of use and engagement.



User - 3
"The new interface is fantastic! It’s intuitive and easy to navigate, with all the information I need right where I expect it to be. The design makes completing tasks effortless. Great job on creating such a user-friendly experience! "
User - 3
"The new interface is fantastic! It’s intuitive and easy to navigate, with all the information I need right where I expect it to be. The design makes completing tasks effortless. Great job on creating such a user-friendly experience! "
User - 4
"The mascot looks cute and friendly"
User - 4
"The mascot looks cute and friendly"
Explore the prototype here
Results












Learnings
From the Arizona Water Chatbot project, I learned that user feedback is vital for creating technology solutions that truly meet user needs. Challenges like balancing diverse stakeholder priorities and maintaining ASU's brand while using a different mascot taught me the importance of adaptability and creative problem-solving. Detailed usability testing and competitive analysis were key in refining the product, while the challenges of real-time data integration underscored the need for seamless, personalized user experiences. Additionally, I realized that crafting natural, intuitive dialogues in the chatbot significantly boosts user engagement and satisfaction.
Under the guidance of Professor Dr. Claire Lauer, I had the opportunity to serve as a Graduate Service Assistant Designer for the Arizona Water Innovation Initiative at the Julie Wrigley Global Institute of Sustainability at ASU. During this time, I contributed to designing games, updated mascots, and participated in various design and research initiatives.
Learnings
From the Arizona Water Chatbot project, I learned that user feedback is vital for creating technology solutions that truly meet user needs. Challenges like balancing diverse stakeholder priorities and maintaining ASU's brand while using a different mascot taught me the importance of adaptability and creative problem-solving. Detailed usability testing and competitive analysis were key in refining the product, while the challenges of real-time data integration underscored the need for seamless, personalized user experiences. Additionally, I realized that crafting natural, intuitive dialogues in the chatbot significantly boosts user engagement and satisfaction.
Under the guidance of Professor Dr. Claire Lauer, I had the opportunity to serve as a Graduate Service Assistant Designer for the Arizona Water Innovation Initiative at the Julie Wrigley Global Institute of Sustainability at ASU. During this time, I contributed to designing games, updated mascots, and participated in various design and research initiatives.
Future Scope
The Arizona Water Chatbot has been successfully developed and launched to the public, demonstrating its potential to revolutionize the way users access and interact with water-related information. Its debut at numerous conferences has showcased how AI-driven chatbots can positively impact lives by providing timely and accurate information, thus enhancing user engagement and satisfaction.
Marketing and Adoption
The chatbot has been actively marketed to ensure widespread adoption, emphasizing its benefits in delivering efficient and personalized water information. This marketing effort aims to reach a broad audience, encouraging more users, businesses, and Government officials to integrate the chatbot into their daily routines.
The further expansion of the project would be educating people about how to save water by designing games and public kiosks at Phoenix Museum for the people to experience the Water Chatbo along with introducing Voice to Search, Making the chatbot multilingual since the state of Arizona has people talking in Spanish and an advanced analytical visualization data.
Future Scope
The Arizona Water Chatbot has been successfully developed and launched to the public, demonstrating its potential to revolutionize the way users access and interact with water-related information. Its debut at numerous conferences has showcased how AI-driven chatbots can positively impact lives by providing timely and accurate information, thus enhancing user engagement and satisfaction.
Marketing and Adoption
The chatbot has been actively marketed to ensure widespread adoption, emphasizing its benefits in delivering efficient and personalized water information. This marketing effort aims to reach a broad audience, encouraging more users, businesses, and Government officials to integrate the chatbot into their daily routines.
The further expansion of the project would be educating people about how to save water by designing games and public kiosks at Phoenix Museum for the people to experience the Water Chatbo along with introducing Voice to Search, Making the chatbot multilingual since the state of Arizona has people talking in Spanish and an advanced analytical visualization data.